A wide range operation power system stabilizer design with neural networks using power flow characteristics

The paper presents a neural network-based power system stabilizer (neuro-PSS) design methodology for a generator connected to a multimachine power system on the basis of the nonlinear power flow dynamics. By utilizing the power flow dynamics, it has not only a wide operation range compared with a conventional PSS design based on a linear model, but also reduces the size of network to train. The neuro-PSS consists of two neural networks: neuroidentifier; and neurocontroller. The rotor dynamics of a generator during low frequency oscillations is modeled by the neuroidentifier with the power flow dynamics. A generalized backpropagation thorough time (GBTT) algorithm is then developed to train the neuro-PSS. The simulation results show that the neuro-PSS designed in this paper performs well with good damping for a wide operation range compared with the conventional PSS